5 ms·
Cooling computer servers with oil
- ktizo 15y ago"Like water, oil is able to absorb more heat — compared to the same volume of air — from the hardware before heating up itself" Am I just being picky, or should I expect better phrased physics from a hardware blog?
- sp332 15y agoI think a cursory description of the concept of specific heat might be relevant to that article. https://en.wikipedia.org/wiki/Specific_heat https://en.wikipedia.org/wiki/Specific_heat But I think it's misleading to say that oil and water have similar heat capacity. Water has an extremely high heat capacity that light oils can't rival.
- ktizo 15y agoWas the "before heating up itself" that rankled, read to me as though they were either confusing the concept of heat capacity with latent heat, or were just wording things badly.
- stcredzero 15y agoAm I just being picky, or should I expect better phrased physics from a hardware blog? I think you should just expect better edited blogs than extremetech.
- mrsebastian 15y agoHey! I've updated the story now, to a slightly more scientific description (involving specific heat capacity). Sorry for any rankling caused :)
- ktizo 15y agoWow, Hacker News appears to be one of the few places where winging on an internet messageboard actually seems to have some effect. I'm going to start complaining that world-peace was promised ages ago and see if that helps any.
- mrsebastian 15y agoWell, feedback is always welcome! The main difficulty is trying to keep track of all that feedback, all over the web. I have some Google Alert things, for mentions of ExtremeTech, and I follow sites like Reddit and HN. It's pretty hard to get valuable feedback, to be honest. Open commenting systems like Disqus definitely lower the barrier to entry, which is nice -- but still, only a tiny fraction of readers bother to comment if something irks/pleases them.
- ktizo 15y agoWell, thanks for sorting that out. Have you got some spare time to get the world peace thing sorted as well, or is that not your department? Oh, and while you're at it, I'd like some donuts and a banana daiquiri. ;)
- mrsebastian 15y agoI'm a writer -- you think I have enough money to buy donuts and alcohol? :P I don't think world peace will ever be possible, not while tech blogs exist.
- ktizo 15y agoOk, in that case I'll settle for some stale bread and a bowl of thin gruel. Don't worry about the world peace thing either, I'll just take more afternoon naps instead.
- nathanb 15y agoWorks great until you have to pull the server and swap out a PCI card....
- jjcm 15y agoKeep latex gloves around. It's easier than you'd think to deal with oil comps.
- freehunter 15y agoSeems like pulling out an oil-drenched card would let oil into the PCI slot, which would then cause issues when a new card is inserted. Do you have experience with that?
- angersock 15y agoI'd imagine just having a compressed air line run over to the workbench would be sufficient. Yank chassis, pull card, blow out oil, replace card, replace chassis.
- pavelkaroukin 15y agowhy this should cause any issues?
- freehunter 15y agoWell oil is non-conductive, so if there's a layer of oil covering the connectors, there might be a performance impact.
- srl 15y agoI have a friend (in college) whose desktop computer is similar: a few circuit boards (motherboard+video, I think) stuck in a bucket of oil, with a radiator+pump, monitor, keyboard, and mouse attached. Apparently, it's effective, although it's difficult to suppress the urge to tip the thing over. It's probably not cost-effective as far as power usage, but that's not his concern - he just didn't want fans.
- K2h 15y agoHow about a hard disk? how does that do submerged? Maybe there is a business opportunity there. There was no talk of reclaiming the waste heat to do anything useful with - I image starting with heat in oil may be easier to do cool things with than pulling it directly out of air.
- dasil003 15y agoAren't HDDs airtight due to the risk of dust crashing the disk?
- DanBC 15y agoNo. They have a small hole, with a version of goretex. This is to deal with different air-pressure. The heads ride on a cushion of air, so it's pretty important not to mess with that. There's usually a small silica gel bag inside the case too, to cope with moisture. Finally, Google did research showing that hot hard drives are not really a problem; they're not more likely to fail just because they're hot. (Until they get very hot, I think.)
- mistercow 15y agoApparently harddrives are not completely hermetically sealed for pressure equalization reasons, so you'd have to do some fiddling to submerge them. I'm not clear what the point of submerging the harddrive is though.
- freehunter 15y agoYeah, hard drives themselves don't generate a lot of heat. Most heat issues with hard drives come from extra heat in the case not being efficiently moved out. If it's just a couple hard drives by themselves, passive air cooling should be enough to keep them running.